Saudi Arabia's OmniOps Launches Bunyan Inference Hub for In-Kingdom AI Model Access

The product is aimed at organisations that want to put AI applications into production but must keep their data and workloads inside the Kingdom. OmniOps said such organisations often end up building and running their own infrastructure, or juggling several model providers and computing environments, to meet data residency and security requirements. Running AI models, or inference, is becoming the largest recurring AI cost for companies that have moved beyond pilots. In-country hosting is a central part of Saudi Arabia's push to build its own AI capacity.

Customers can choose from a catalogue of models or bring their own. They buy predefined token bundles across the supported models and workloads, and organisations that need more capacity or isolation can rent dedicated GPU infrastructure. The company did not give prices, a list of supported models or launch customers.

OmniOps said the platform runs on its own inference optimisation engine, built by its research and development team. The engine adjusts how models run on the available hardware to match each workload, which the company said improves GPU utilisation, lowers the cost per token and keeps performance steady as demand grows. The hub works across different GPUs and AI accelerators rather than relying on one chip supplier. It routes each request to a suitable model and computing resource, handles scaling, and monitors performance. Administrators can set usage limits and track token consumption by team.

"As entities embrace AI across functions and applications in their operations, ensuring that the consumption of AI models is managed efficiently and with the correct security and cost guardrails has never been more critical," said Mohammed Altassan, OmniOps' founding chief executive. He said the hub packages the company's Bunyan inference technology and its experience in sovereign AI deployment into a platform that public and private sector organisations can plug into their operations.

The new hub builds on Bunyan, the inference-as-a-service platform OmniOps launched in July 2025 and described at the time as the Kingdom's first sovereign service of its kind. The company, founded in 2024, has also signed a memorandum of understanding with US inference chip company Groq to explore joint opportunities.

In its announcement, OmniOps cited ServiceNow research showing that Saudi-based organisations raised AI spending by 124 per cent year on year and plan to put 20 per cent of their IT budgets into AI until 2027. It also cited a Gartner forecast that global spending on AI inference workloads will reach $23.3 billion by the end of 2026.

OmniOps did not say when the hub will be generally available or what it will charge per token. Those details will decide whether it can compete with larger cloud providers that are also adding in-Kingdom capacity.